Data Quality Observation in Pervasive Environments

Pervasive applications are based on acquisition and consumption of real-time data from various environments. The quality of such data fluctuates constantly because of the dynamic nature of pervasive environments. Although data quality has notable impact on applications, little has been done on handling data quality in such environments. On the one hand past data quality research is mostly in the scope of database applications. On the other hand the work on Quality of Context still lacks feasibility in practice, thus has not yet been adopted by most context-aware systems. This paper proposes three metric definitions - Currency, Availability and Validity - for pervasive applications to quantitatively observe the quality of real-time data and data sources. Compared to previous work, the definitions ensure that all the parameters are interpretable and obtainable. Furthermore, the paper demonstrates the feasibility of proposed metrics by applying them to real-world data sources on open IoT platform Cosm (formerly Pachube).